Trace Interviews Step-By-Step

In this penultimate post for The Person in the (Big) Data edition of EM, Elizabeth Dubois @lizdubois provides
a step-by-step account of how the trace interviewing process works. Trace interviewing is a method used to elicit a person’s stories about why they made particular traces on social media platforms and is a wonderful way of getting at the stories underlying the data. Elizabeth’s step-by-step will hopefully help others wanting to use a similar method in the interview process! 

Social media data and other digital traces we leave as we navigate the web offer incredible opportunity for discovery within the social sciences. I am going to take you step by step through the process of trace interviewing – an approach that helps researchers gain richly detailed insights about the social context of that digital data. For those interested in the why as well as the how, Heather Ford and I talk a lot about why we think the method is important and what types of things it can offer researchers (such as validity checks, information about social context, opportunities to join data sets from various platforms) in our paper.

The Study
I want to figure out how people who influence other people on political issues choose their channels of communication (and the impact of those choices). The only way to understand these decisions and their impacts, I think, is a mixed-methods approach. So, I draw on Twitter data for content and network analysis, an online survey and in-depth trace interviews. You can read more about the full work here.

Trace Interviews
Hey, so great to finally meet you in person. Welcome!
By the time I got to the interview stage my interviewee and I already knew quite a lot about each other. They had filled out a survey, they knew I found them because the use the #CDNpoli hashtag, and they had read a project description and signed a consent form in advance.

It was important to form a relationship with my participants well in advance because I needed permission to collect their data. Sometimes trace data is publicly available (for example, tweets made or the list of accounts a Twitter user follows). But, even when it is publicly available, I tend to think that giving your participant a heads up that you’ve got or will collect data specifically about them is a good call. The fact is, people don’t always understand what making a public tweet means.

For my study I also collected Facebook data, Instagram data and Reddit data where available so I needed consent and specific access. The average participant actively used Facebook and Twitter so those are the two I talk about below.

I’d like to start by getting a baseline of your social network – I mean your everyday network that includes on and offline interaction.

It is important for me to be able to understand how and when social media are used as well as other channels of communication such as telephone, text messaging, email and even face-to-face conversations. In order to do this I modified the participant-aided sociogram technique (Hogan, Carassco and Wellman, 2007).

A participant placing tags on their sociogram

Below are the basic instructions for participants:

1. Here are a bunch of post-it notes. List out everyone who you consider to be “very close” and “somewhat close”.
2. Here is a legend with numbers indicating your relationship with each person. Add a little number to each tag.
3. Here is a big piece of paper, imagine you are in the centre of this set of 4 circles, place each of those on the list you just made on circles. Close to you on the paper means close to you socially and the farther away they get on the paper, the less close you feel to them.
4. You want to arrange (and re-arrange) those tags until the people who know each other are all in the same area of the paper. You want to be creating little clusters – think of the lunch tables in the cafeteria in high school.
5. Describe each group and who knows who. Draw lines and circles to show relationships.

My addition is at step 2. I gave each participant a list of numbers but I also gave them a handful of markers. I asked them to put a blue dot next to anyone they are connected to on Twitter, a red dot for Facebook friends, a pink one for Instagram (etc.). I also asked them to put a brown star next to anyone with whom they’d had a conversation about politics in the past six months.

The image below outlines the core pieces of information I gathered at this stage. A: the cliques/friend groups. B: the relationship (noted in step 2). C: the channels of communication. D: overlap with social media networks (in this case the participant’s Facebook network).

Information gathered using the participant-aided sociogram

Base established, I could now ask questions about what kinds of people the participants communicated with using various channels. I could also overlay the Facebook and Twitter networks I’d collected and find gaps or anomalies. Importantly, through the physical process of crafting their own social network on paper, I had started to teach my participants how to read the online social network visualizations I was about to present them with.

This is your Facebook ego-network, it shows all your friends. A dot is a friend (we call them nodes) and a line is a friendship (we call them edges).

Participants had normally logged on to Facebook and authorized an app called “NameGenWeb” (no longer available) before the interview at my request. We’d open it up together and mouse over different clusters and nodes to identify groups. We’d work together to explain how the groups were similar or different from the sociogram created in the first half of the interview. I’d ask about individuals who were prominent in the sociogram and this would often lead to interesting discussions about why they do or do not use Facebook with one friend over another.

We repeated the process for their Twitter mention network which I had collected for a two-week period prior to the interview using Netlytic. This kind of network is very different from a Facebook friend network and so provided another perspective on the interviewee’s political communication practices. It was conversation in the interview setting that allowed me to understand each network in the larger social context of my participant’s daily life.

Example of a Facebook friendship network using NameGenWeb

An example of a Twitter mention network using Netlytic

This has been great! I am wondering if you think a few of the people we’ve talked about today might be willing to chat?

Because I was interested in influence I also went on to conduct short and sweet interviews with some of the people identified during the trace interview process. Using an in-depth (and thus often time consuming) method like trace interviews with key informants can help you better identify other people, events or even hashtags to investigate.

Quick Reflections
My approach combined a lot of different kinds of data and took 2-3 hours in person but not every trace interview has to be like that. For example, Heather Ford conducted short trace interviews over Skype with Wikipedia editors to better understand how they chose which sources to cite when editing articles about current affairs.

I’ve only had time to skim the surface here but believe that frank discussions about how we can (dare I say, should) integrate digital data into interviews is important. I’d love to chat more so feel free to comment below or find me on Twitter: @lizdubois

Translation

English中文(简体)FrançaisEspañol

by

You can help improve the automatic translation by checking off “Edit Translation” after you select a language. Thanks!